跌倒在精神科住院病人的發生率是相當高的,但缺乏有效驗證之精神科病人跌倒評估量表,以輔助醫療人員作正確判斷,故本研究目的為: 1.比較精神科住院病人跌倒風險評估量表(Psychiatric Inpatient Fall Risk Assessment Tool, PIFRAT)、Wilson-Sims Fall Risk Assessment Tool (WSFRAT)之敏感度、特異度、準確度、陽性預測值、陰性預測值之差異。2.預測PIFRAT及WSFRAT於精神科科住院病人之跌倒危險因素。3.比較精神科護理人員使用PIFRAT及WSFRAT之滿意度差異。本研究為診斷性試驗研究(diagnostic test studies),資料收集自 2016.10.01起至2017.3.10止,於台灣北部某精神專科醫院進行入院新病人住院第1-7天之跌倒評估資料收集與護理師對使用跌倒風險評估量表之滿意度問卷調查。資料採用描述性統計、邏輯斯回歸分析、信效度檢測、工具效能檢測及ROC曲線等進行資料分析。 本研究共收案234位,住院期間跌倒者有28位(12%),無跌倒206位(88%)。本研究結果顯示兩種跌倒評估量表之敏感度都不佳(WSFRAT 57.1%, PIFRAT 50%);但特異度WSFRAT (79.6%) 優於PIFRAT( 70.4%);準確度WSFRAT( 90.9%)較PIFRAT (80.3%)佳。然而,兩種評估量表有較低的PPV值(WSFRAT=27.6%,PIFRAT=18.7%)和高的NPV值高(WSFRAT=93.2%,PIFRAT=91.2%)。ROC曲線分析顯示,PIFRAT之 AUC= 0.602,根據Youden指數最適合的切點為1.5分可達96.4%的敏感度; WSFRAT之 AUC= 0.625,根據Youden指數最佳切點為3.5分可達82.1%的敏感度。多變量邏輯斯迴歸分析顯示,過去有跌倒經驗者再發生跌倒的風險顯著高於沒有跌倒經驗者;而性別(OR=2.57, CI: 1.11-5.94)、身體活動困難(OR= 3.43 ; CI: 1.40-8.41)、虛弱無力(OR=3.03; CI: 1.08-8.49)等跌倒相關因素亦達統計上之顯著。在滿意度結果方面,護理人員使用PIFRAT之滿意度總分47.9分(SD =6.9)與WSFRAT之47.3分(SD =6.4),並無顯著的差異。 本研究結果將可作為未來發展跌倒評估量表之參考,以找出符合本土具預測力的精神科跌倒評估工具。
The incidence of falls is very high in psychiatric inpatients; however, lack of effective validated psychiatric inpatient fall risk assessment tools to assist the medical staff in making a correct judgment. Therefore, the purposes of this study were: 1. to compare the sensitivity, specificity, accuracy, positive predictive value, and negative predictive value between psychiatric inpatient fall risk assessment tool (PIFRAT) and Wilson-Sims fall risk assessment tool (WSFRAT); 2. to predict falls risk factors for PIFRAT and WSFRAT in psychiatric inpatients; 3. to compare the satisfaction level between PIFRAT and WSFRAT in psychiatric nurses. This study was diagnostic test studies. Data were collected from 2016/10/01 to 2017/03/10. Fall assessment data were collected from new patients admitted to psychiatry units at 1st to 7th days in northern Taiwan. Nurses’ satisfaction related to fall risk assessment tool was surveyed. Data were analyzed using descriptive analysis, logistic regression analysis, reliability and validity testing, tool effective testing, and receiver operating characteristic (ROC) curve analysis. We recruited 234 cases in this study, 28 (12%) fallers and 206 (88%) nonfallers during data collection period. The results revealed that these two fall risk assessment tools had low sensitivity (WSFRAT 57.1%,PIFRAT 50%), but the specificity of WSFRAT (79.6%) was higher than PIFRAT (70.4%), and the accuracy of WSFRAT (90.9%) was also higher than PIFRAT (80.3%). Nevertheless, these two assessment tools had lower PPV values (WSFRAT = 27.6%, PIFRAT = 18.7%) and high NPV values (WSFRAT = 93.2%, PIFRAT = 91.2%). The ROC curve analysis revealed that PIFRAT AUC was 0.602. According to the Youden index, the best cut-off point was 1.5 points and reached to 96.4% of the sensitivity. WSFRAT AUC was 0.625, the Youden index suggests cut-off point was 3.5 points and reached to 82.1% of the sensitivity. Multivariate logistic regression analysis revealed that patients who have fall history in the past, the risk of falling again was significantly higher than those who haven’t. We found that female (OR = 2.57, CI: 1.11-5.94), physical activity disturbance (OR=3.43; CI:1.40-8.41), weakness(OR=3.03; CI:1.08-8.49) were significantly associated with fall. There was no significant difference in nurses’ satisfaction level: PIFRAT 47.9 points (SD = 6.9) and WSFRAT 47.3 points (SD = 6.4). The findings can serve as references for future development of fall assessment tools in order to establish a local psychiatric fall risk assessment tool.